Multiobjective clustering with automatic k-determination for large-scale data
|
|
- Anastasia Vivien Richards
- 5 years ago
- Views:
Transcription
1 Multiobjective clustering with automatic k-determination for large-scale data Scalable automatic determination of the number of clusters ABSTRACT Web mining - data mining for web data - is a key factor of web technologies. Especially, web behavior mining has attracted a great deal of attention recently. Behavior mining involves analyzing the behavior of users, finding patterns of user behavior, and predicting their subsequent behaviors or interests. Web behavior mining is used in web advertising systems or content recommendation systems. To analyze huge amounts of data, such as web data, data-clustering techniques are usually used. Data clustering is a technique involving the separation of data into groups according to similarity, and is usually used in the first step of data mining. In the present study, we developed a scalable data-clustering algorithm for web mining. To derive clusters, we applied multiobjective clustering with automatic k-determination (MOCK). It has been reported that MOCK shows better performance than k-means, agglutination methods, and other evolutionary clustering algorithms. MOCK can also find the appropriate number of clusters using the information of the trade-off curve. The k-determination scheme of MOCK is powerful and strict. However the computational costs are too high when applied to clustering huge data. In this paper, we propose a scalable automatic k-determination scheme. The proposed scheme reduces Pareto-size and the appropriate number of clusters can usually be determined. Keywords Multi-objective optimization, Pattern recognition and classification, Data mining, Speedup technique Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Copyright 200X ACM X-XXXXX-XX-X/XX/XX...$ INTRODUCTION We are studying data clustering for web behavior analysis. Data clustering is a technique of separating huge datasets into groups according to their similarities. Web behavior analysis is performed to find characteristic behavior patterns of users and predict their subsequent behaviors. Web behavior analysis is the key factor behind personalization and effective ad delivery systems, etc. This clustering technique is very important to allow the analysis of millions of sets of web behavior data. There are many techniques for data clustering, like k means, the agglutination method, and other methods based on EAs, etc[3][4][5]. The main points involved in clustering are determining the appropriate number of clusters 1, and finding the partitions among k clusters. k-means is better at finding partitions but worse at determination of appropriate k, while, the agglutination method is better at determination of appropriate k but worse at finding partitions. Each method uses a single objective, the objective of k-means is compactness of clusters and that of agglutination is connectedness of similar data. In our study, we used a multiobjective evolutionary clustering algorithm with these two objectives. We used the multiobjective clustering algorithm proposed by Handl and Knowles[1][2], MOCK (Multiobjective clustering with automatic k determination)[1]. This algorithm optimizes two complementary objectives based on cluster compactness and connectedness. It has been reported that MOCK finds the partitions better than k-means, agglutination methods, and other evolutionary clustering algorithms. It also has an automatic k-determination scheme and simultaneously finds the appropriate number of clusters and partitions among the k clusters. MOCK is one of the most powerful and interesting algorithms for data-clustering problems. One goal of our research was to apply MOCK to web data clustering. However, the original algorithm is difficult to use for web data, because of the limits of the computation cost. Especially, the cost of the automatic k-determination scheme is huge. Therefore we investigated a scalable automatic k-determination scheme. In this paper, we propose the scalable automatic k-determination scheme. The remainder of this paper is organized as follows. Section II presents a review of MOCK. In Section III, we describe the scalability of the original MOCK algorithm and our new scheme. The effectiveness of our scheme is discussed in Section IV. Finally, Section V concludes this paper. 1 usually called k
2 2. MULTIOBJECTIVE CLUSTERING WITH AUTOMATIC K-DETERMINATION MOCK (Multiobjective clustering with automatic k-determination) is a powerful clustering algorithm based on a multiobjective genetic algorithm. It optimizes two complementary objectives based on cluster compactness and connectedness. One is intended for determination of appropriate k, and the other is to find partitions between k clusters. In this evolutionary multiobjective clustering algorithm, Pareto solutions are a set of different trade-off partitioning over a range of different cluster numbers, k. This algorithm also is able to determine the final solution with good accuracy and diversity. In this section, we describe the MOCK algorithm, graphbased representation of partitioning, initialization using a minimum spanning tree (MST), and two objective functions. The representation allows treating solutions, which contain different numbers of clusters, and the MST-based initialization makes well-connected initial solutions like those made by the agglutination method. After optimizing two objective functions through genetic operations - crossover, mutation, and selection - Pareto solutions with a range of different k are made. Finally, we describe Pareto of MOCK, and the automatic k-determination scheme. The Pareto solutions contain different trade-off partitioning over a range of different k, and the automatic k determination scheme is able to determine the final solution from the Pareto solutions, and find the appropriate k. 2.1 Graph-based representation MOCK uses graph-based representation[6] in which each datum is represented as a node, and an edge between two nodes indicates that these data are in the same cluster. This allows treatment of a wide range of k in a single trial. In Figure 1, each datum is numbered i {1..9}. The dotted lines indicate removed edges. Removing the edges from 8 to 3 and from 6 to 8 separates the nodes into three clusters. All k solutions are represented in this way. Removing edges is the key to creating clusters. It is effective in the initialization scheme with a minimum spanning tree. A gene is represented as the list of edges. The phenotype is the list of edges in the directed graph, and the genotype is the list of node ids to which each edge is connected. In this representation scheme, each node has only one edge connecting from it. 2.2 Initialization using minimum spanning tree A minimum spanning tree (MST) is the shortest tree that contains all nodes in the graph and has no loops. The shortest tree means that the total cost of all edges is the minimum. In the MST created for a given dataset, similar data are connected and dissimilar data aren t. The MST of Figure 1 is shown at the top of Figure 2. Removing k 1 edges results in the creation of a k cluster. The removed edges are selected according to their cost; the edges with higher cost are removed. In this way, MOCK makes highquality initial solutions 2. (See bottom of Figure 2) MOCK uses Prim s algorithm[8] to create the MST. This algorithm has order N log N time complexity. 2.3 Objective functions 2 The initial solutions of MOCK are similar to those created by the agglutination method. Figure 1: Graph-based representation Figure 2: MST-based initialization
3 The two objectives of MOCK are connectivity and overall deviation. Connectivity is based on cluster connectedness, and overall deviation is based on cluster compactness Connectivity Connectivity is intended to divide similar data into the same cluster. It is defined as equation (1), where i is the data id, C is the total dataset, C k is the kth cluster, and nn i(j) is the ordinal number 3. Conn(C) = NX LX i=1 j=1 x i,nni (j), where x r,s = ( 1 j if C k : r, s C k 0 otherwise (1) Figure 3: Pareto of MOCK Connectivity is minimized. When all similardata 4 are clustered in the same cluster, connectivity of the solution becomes zero. When these data are in different clusters, connectivity increases 1/j. Therefore, minimizing connectivity decreases k. When k = 0, connectivity is always zero Overall deviation Overall deviation, defined as equation (2), is intended to make clusters more compact. δ() is the distance function designed to calculate the similarity between every datum in the dataset. µ k is the center coordinate or the median datum of kth cluster. Dev(C) = X C k C X i C k δ(i, µ k ) (2) Overall deviation is also minimized. More compact separation makes this value smaller. Therefore, minimizing overall deviation increases the number of clusters. When separating N data into N clusters, overall deviation is zero. Minimizing these two objectives, a wide range of different k solutions are generated. This is efficient to find the appropriate number of clusters, k. For each k, overall deviation is minimized and most compact clusters are made. Therefore, MOCK returns a range of different k solutions, and each solution is well separated. 2.4 Pareto solutions While reduction of overall deviation, each cluster size becomes smaller and k becomes larger. In contrast, with reducing connectivity, each cluster size becomes larger because many neighbor data are collapsed into same clusters, and k become smaller. Therefore, Pareto of MOCK contains solutions over a wide range of k, as shown in Figure 3. In Figure 3, smaller k solutions are on the top left and more k solutions are on the bottom right. The Pareto curve shows the relation between k and δd/δc, where C is connectivity and D is overall deviation. This relation is used in the next step, the automatic k-determination scheme. 2.5 Automatic k-determination scheme 3 nn i (j) is the jth nearest data from i 4 Data within Lth nearest each datum. The automatic k-determination scheme of MOCK is based on the Gap statistic[7]. In this scheme, we make random data 5 based on principal component of the original data, and divide these data into clusters by MOCK. The Pareto curves of the random and original data are different. In the original Pareto, the most distant solution from the nearest solution of all Control solutions is selected as the most characteristic solution. In the Gap statistic, we guess that such a solution has appropriate k. (Figure 4) Figure 4: MOCK Automatic k-determination scheme of As described above, the k-determination scheme of MOCK requires a random data clustering step. The calculation cost is as high as when clustering the provided dataset. If using the random data clustering step in M trial, the original calculation cost is M + 1 times as large. This computational cost is very high, especially in large-scale dataset such as web data. Therefore, we propose a scalable k-determination scheme without the random data clustering step in the next section. 5 called Control data
4 3. IMPROVEMENT OF SCALABILITY OF THE K-DETERMINATION SCHEME Determination of the number of clusters is one of the most important concerns in the clustering problem. In this process, the k-determination scheme of MOCK based on the GAP statistic works well. However, as mentioned in the previous section, the original scheme has a problem. The computation costs for determining the cluster number in the original MOCK scheme are high. The cost is serious problem when classifying over a million data. In this section, we describe the cost of the original k- determination scheme. In addition, we propose new scalable k-determination scheme, which uses the information of the Pareto curve. In our scheme, the relation between connectivity and overall deviation is very important. 3.1 Relation of connectivity and overall deviation In Figure 3, fewer k solutions are on the top left, and more k solutions are on the bottom right. We can consider the Pareto curve as a function of k. When Pareto solutions contain the solution of the appropriate k, the smaller k solutions are on the left of the appropriate k solution, and larger on the right. The left figure in Figure 5 shows the situation where the data should be separated into two clusters. In this situation, when the cluster is separated into two, the value of connectivity is small and the overall deviation becomes smaller. On the other hand, the figure on the right shows the situation where the data should not be separated. In this situation, when the cluster is separated into two, the overall deviation becomes smaller but the value of connectivity increases. This is because the number of data penalized becomes very large. Therefore, δd/δc, where C is connectivity and D is overall deviation, rapidly decreases near the appropriate k solution. This knee[9] is shown in Figure 4. reduction step helps to determine the appropriate k more easily in the next step. The second step is also composed of two sub-schemes. The first scheme usually determines the appropriate solution, although it makes errors in several situations. The second scheme overcomes the defect of the first. This set of k-determination sub-schemes is applied in parallel. Each scheme is applied to same reduced Pareto solutions, and returns each final solution. When the returned solutions are not the same, the user chooses either one of the two. Start Initialization GA operations Terminal criterion K-determination End Pareto reduction K-determination by 1st scheme K-determination by 2nd scheme Figure 6: The flow of our automatic k-determination scheme Reduction step The set of Pareto solutions obtained by the original MOCK algorithm contains several same-k solutions, and low-potential solutions. These solutions are removed in this step. (Figure 7) Figure 5: Should be separated or not Figure 7: Reduction of Pareto size 3.2 Scalable k-determination scheme Figure 6 shows the operation flow of MOCK and our new k-determination scheme. Our proposed scheme is composed of two steps: a Pareto-size reduction step and a k- determination step. In the first step, removing the local non-convex part of solutions reduces the Pareto size. This We empirically assume that Pareto is convex. The criterion to remove is that the solution is in a partly non-convex part. Solutions in the partly non-convex part do not have more compact clusters than those of the nearby convex part. Figure 8 shows the convex and non-convex sample solutions in Pareto of Figure 7. S1-S3 are part of the Pareto solutions returned by MOCK, and all these solutions are obtained
5 when k = 3. S1 and S3 are in a partly non-convex part of Pareto and S2 is in a convex part. Figure 9: Failure pattern of the first sub-scheme Figure 8: Non-convex and convex part of solutions In S1, almost all data are divided into the two largest clusters, and a few data are divided into the smallest cluster. This solution is similar to those obtained when k=2. Such a solution is worse than the k=2 solutions on the left, because the smallest cluster contains too few data and has little information. Solutions S2 and S3 seem to be the same. However, connectivity is not the same, so the border of the above two clusters is more efficient in S2 than in S3. Thus, S1 and S2 are lower-potential solutions than S2. The flow of this step is as follows. 1. Normalize overall deviation and connectivity to [0, 1] - [1, 0]. 2. Sort solutions by connectivity. 3. Calculate adjacent angles. When it is non-convex, remove the solution. This step is iterated until no more solutions are removed k-determination step: first scheme Our k-determination scheme uses the information of the Pareto curve, δd/δc. We determine the solution on the knee of the Pareto curve as the final solution. This is not as powerful as the original MOCK scheme 6. However, this is a very simple and efficient method to determine the appropriate k, which does not require the random data clustering step. The process is described below. 1. Reduction step. 2. Calculate adjacent angles and determine the knee k-determination step: second scheme Figure 9 shows the failure pattern of the first sub-scheme. This dataset is composed of non-circular clusters. 6 This scheme makes errors when the typical knee is not shown in the Pareto, although the reduced Pareto contains the appropriate k solution. Such non-circular clusters foil the first sub-scheme for the following two reasons. 1. Overall deviation decreases any time separating clusters, and circular clusters are more efficient than noncircular ones using the Euclidean distance. 2. When cutting such a wide cluster, penalized data in connectivity function are fewer than circular clusters. The following steps are our basic solution to solve this problem. When separating a wide non-circular cluster, keep δd/δc the same as when it is circular. This adjustment is done in overall deviation function, because the connectivity function uses ordinal numbers and it is difficult to change the value. We accept the scaling filter to the overall deviation function, after obtaining the Pareto solutions. The flow below is our filter method. 1. Reduction step. 2. Make each cluster of each solution non-correlated by PCA. 3. Calculate variance of each dimension of each cluster, and scale down each dimension to the minimum-variance dimension. 4. Re-valuate overall deviation of each solution, and create new Pareto. 5. Determine the new knee. We can treat non-circular clusters in this sub-scheme.
6 4. ASSESSMENT EXPERIMENT 4.1 Proposal of the experiment This experiment was performed to investigate the performance of our new scheme. For this purpose, we used 9 characteristic test datasets: basic circulate clusters, overlapping clusters, non-circular clusters, circular and non-circular clusters, different distribution in clusters. We applied our first and second sub-scheme to each dataset, and investigated whether the appropriate k solution was selected by each scheme, and the size of the reduced Pareto. Notice that, in all test datasets used, the appropriate k solution was always created by MOCK and these were in the Pareto. 4.2 Dataset We used 9 test datasets employed in [1]. Their names, size 7 and appropriate k are listed in Table 1, and Figure 10 shows an overview. They are all 2D continuous-valued datasets. Table 1: Test dataset Dataset Size Appropriatenumberofclusters Square Square Sizes Triangle Long Twenty D2C Spiral LongSquare Square1 and Square4 consist of four circular clusters of equal size and spread. They differ in the overlapping of clusters. The sizes of clusters in Sizes5 are not equal. Triangle1 consists of a non-circular cluster. Long1 consists of only non-circular clusters. Twenty has twenty clusters of equal size, circularity, and spread. D2C10 consists of ten noncircular and circular clusters of different size and spread. LongSquare consists of Long1 and Square Parameter settings The parameters used in this experiment are listed in Table 2. These are recommended values for MOCK. (See [10]) In our MOCK, the internal population size is equal to the external population size, because we used SPEA2[11] as the base multiobjective genetic algorithm of MOCK. Handl s MOCK is based on PESA-II[12]. We also used archive copy selection and neighborhood crossover schemes, proposed by Watanabe[13], as the internal population selection scheme and for crossover of SPEA2, respectively. 4.4 Experimental results We investigated the performance of auto k-determination and reduction rate of Pareto size when applying our proposed scheme to MOCK. Figure 11 shows the k of the selected solution by our two sub-schemes. The thick line is the appropriate k for the dataset, which is described in Table 1. In Square1, Square4, 7 i.e., number of data Table 2: Parameter settings for MOCK P arameter Setting Gene length(n) dataset size Number of generations 200 Internal population size min(50, 2*k max ) External population size min(50, 2*k max ) Initialization Minimum Spanning Tree Crossover Uniform and neighborhood crossover Crossover rate 0.7 Mutation type L nearest neighbors(l = 20) Mutation rate 1/N + (l/n) 2 Constraints k {1,.., 25}, cluster size>2 Sizes5, Triangle1, and Twenty, the first scheme found the appropriate k perfectly. However, in Long1 and Spiral, the first scheme did not find appropriate k. In these cases, the second sub-scheme covered the first. In this result, either of our two sub-schemes can find the appropriate k. Comparison of this result and [1] indicated that the performance of our scheme is worse than that of the original scheme. However, our scheme can be used without random data clustering, which can reduce the calculation costs over 50% as compared with the original MOCK. This is one of the most important factors in web mining. Table 3 shows the reduced Pareto size (average of 10 trials). Over the whole dataset, about 70% of solutions are removed in the first sub-scheme, and 50% more are removed in the second. In addition, in the case in which the auto k-determination scheme makes a mistake in identifying k, the appropriate k solutions remained in the reduced Pareto. This result indicated that our scheme reduced the Pareto size by about 50-70%. It is efficient in cases in which an expert selects the last solution when they are not satisfied with the selected solution. Table 3: Reduction rate of Pareto size Dataset 1stscheme 2ndscheme Square1 76.4% 60.8% Square4 77.8% 49.2% Sizes5 74.2% 60.4% Triangle1 73.0% 74.6% Long1 70.4% 69.6% Twenty 84.7% 72.5% D2C % 57.6% Spiral 74.4% 74.2% LongSquare 68.0% 58.8%
7 Figure 10: Test dataset overview Figure 11: Selected k in each dataset
8 5. CONCLUSIONS Data clustering is an important technique used in the preparation step of web mining. Many methods are available to do this, each of which has both merits and demerits. We used a multiobjective clustering algorithm, MOCK, for clustering performance. It has been reported that MOCK is better than k-means and agglutination methods. However, the calculation costs associated with this algorithm are too high for use with large-scale data, such as web access data. Especially, automatic k-determination scheme has high calculation cost. We attempted to resolve this problem. We proposed a new automatic k-determination scheme of MOCK, which includes two sub-schemes. The first subscheme using the information of the Pareto curve is effective but produces fatal errors under some conditions. The second sub-scheme covers this drawback. Our new scheme is able to determine the appropriate k at low cost, although the performance is poorer than the original algorithm. In addition, our scheme can reduce the Pareto size by about 50-70%. Using this scheme, MOCK can be applied to largescale data. In future work, the proposed schemes will be applied to realistic web data clustering and behavior mining. of the 2005 IEEE Congress on Evolutionary Computation, pages IEEE Press, Anaheim, CA, [11] E. Zitzler and M. Laumanns and L. Thiele. SPEA2: Improving the Performance of the Strength Pareto Evolutionary Algorithm. Tech- nical Report 103, Computer Engineering and Communication Networks Lab(TLK), Swiss Federal Institute of Technology(ETH)Zurich(2001), May [12] David W. Corne, Nick R. Jerram, Joshua D, Knowles, Martin J. Oates. PESA-II: Region-based Selection in Evolutionary Multiobjective Optimization. Proceedings of the Genetic and Evolutionary Computation Conference, [13] S. Watanabe. Genetic Algorithm Based on neighborhood Crossover for Multi-Objective Optimization. Doctoral thesis, Doshisha University, JPN, ADDITIONAL AUTHORS 7. REFERENCES [1] Julia Handl and Joshua Knowles. Multiobjective clustering with automatic determination of the number of clusters, Technical Report No. TR-COMPSYSBIO , UMIST, Department of Chemistry, August [2] Julia Handl and Joshua Knowles. Evolutionary Multiobjective Clustering, in Xin Yao et al. (editors), Parallel Problem Solving from Nature (PPSN VIII), pp , Springer-Verlag, Lecture Notes in Computer Science, Vol. 3242, Berlin, September [3] A.K.Jain, M.N.Murty, and P.J.Flynn. Data clustering: A review. ACM Computing Surveys, 31: , [4] L.Kaufman and P.J.Rousseeuw. Finding Groups in Data. John Wiley & and Sons Inc., New York, NY, [5] Ujjiwal Maulik, Sandhamitra Bandypadhyay. Genetic algorithm-based clustering technique. Pattern Recognition 33, pages , [6] Julia Handl and Jushua Knowles. An investigation of representations and operators for evolutionary data clustering with a variable number of clusters. Parallel Problem Solving from Nature, [7] R. Tibshirani, G. Walther, and T. Hastie. Estimating the number of clusters in a dataset via the Gap statictic. Technical report, Stanford University, [8] R. C. Prim. Shortest connection networks and some generalizations. Bell System Technical Journal, 36: , [9] J. Branke, K. Deb, H. Dierof, and M. Osswald. Finding knees in multi-objective optimization. Proceedings of the Eighth International Conference on Parallel Problem Solving from Nature, Birmingham, UK, [10] Julia Handle, Jusua Knowles. Improvements to the scalability of multiobjective clustering. In Proceedings
NCGA : Neighborhood Cultivation Genetic Algorithm for Multi-Objective Optimization Problems
: Neighborhood Cultivation Genetic Algorithm for Multi-Objective Optimization Problems Shinya Watanabe Graduate School of Engineering, Doshisha University 1-3 Tatara Miyakodani,Kyo-tanabe, Kyoto, 10-031,
More informationComparison of Evolutionary Multiobjective Optimization with Reference Solution-Based Single-Objective Approach
Comparison of Evolutionary Multiobjective Optimization with Reference Solution-Based Single-Objective Approach Hisao Ishibuchi Graduate School of Engineering Osaka Prefecture University Sakai, Osaka 599-853,
More informationDCMOGADES: Distributed Cooperation model of Multi-Objective Genetic Algorithm with Distributed Scheme
: Distributed Cooperation model of Multi-Objective Genetic Algorithm with Distributed Scheme Tamaki Okuda, Tomoyuki HIROYASU, Mitsunori Miki, Jiro Kamiura Shinaya Watanabe Department of Knowledge Engineering,
More informationLamarckian Repair and Darwinian Repair in EMO Algorithms for Multiobjective 0/1 Knapsack Problems
Repair and Repair in EMO Algorithms for Multiobjective 0/ Knapsack Problems Shiori Kaige, Kaname Narukawa, and Hisao Ishibuchi Department of Industrial Engineering, Osaka Prefecture University, - Gakuen-cho,
More informationRecombination of Similar Parents in EMO Algorithms
H. Ishibuchi and K. Narukawa, Recombination of parents in EMO algorithms, Lecture Notes in Computer Science 341: Evolutionary Multi-Criterion Optimization, pp. 265-279, Springer, Berlin, March 25. (Proc.
More informationSPEA2+: Improving the Performance of the Strength Pareto Evolutionary Algorithm 2
SPEA2+: Improving the Performance of the Strength Pareto Evolutionary Algorithm 2 Mifa Kim 1, Tomoyuki Hiroyasu 2, Mitsunori Miki 2, and Shinya Watanabe 3 1 Graduate School, Department of Knowledge Engineering
More informationLate Parallelization and Feedback Approaches for Distributed Computation of Evolutionary Multiobjective Optimization Algorithms
Late arallelization and Feedback Approaches for Distributed Computation of Evolutionary Multiobjective Optimization Algorithms O. Tolga Altinoz Department of Electrical and Electronics Engineering Ankara
More informationA Model-Based Evolutionary Algorithm for Bi-objective Optimization
A Model-Based Evolutionary Algorithm for Bi-objective Optimization Aimin Zhou 1, Qingfu Zhang 1, Yaochu Jin 2, Edward Tsang 1 and Tatsuya Okabe 3 1 Department of Computer Science, University of Essex,
More informationEvolutionary Algorithms: Lecture 4. Department of Cybernetics, CTU Prague.
Evolutionary Algorithms: Lecture 4 Jiří Kubaĺık Department of Cybernetics, CTU Prague http://labe.felk.cvut.cz/~posik/xe33scp/ pmulti-objective Optimization :: Many real-world problems involve multiple
More informationA Similarity-Based Mating Scheme for Evolutionary Multiobjective Optimization
A Similarity-Based Mating Scheme for Evolutionary Multiobjective Optimization Hisao Ishibuchi and Youhei Shibata Department of Industrial Engineering, Osaka Prefecture University, - Gakuen-cho, Sakai,
More informationUsing ɛ-dominance for Hidden and Degenerated Pareto-Fronts
IEEE Symposium Series on Computational Intelligence Using ɛ-dominance for Hidden and Degenerated Pareto-Fronts Heiner Zille Institute of Knowledge and Language Engineering University of Magdeburg, Germany
More informationImproved Pruning of Non-Dominated Solutions Based on Crowding Distance for Bi-Objective Optimization Problems
Improved Pruning of Non-Dominated Solutions Based on Crowding Distance for Bi-Objective Optimization Problems Saku Kukkonen and Kalyanmoy Deb Kanpur Genetic Algorithms Laboratory (KanGAL) Indian Institute
More informationEvolutionary multi-objective algorithm design issues
Evolutionary multi-objective algorithm design issues Karthik Sindhya, PhD Postdoctoral Researcher Industrial Optimization Group Department of Mathematical Information Technology Karthik.sindhya@jyu.fi
More informationFinding a preferred diverse set of Pareto-optimal solutions for a limited number of function calls
Finding a preferred diverse set of Pareto-optimal solutions for a limited number of function calls Florian Siegmund, Amos H.C. Ng Virtual Systems Research Center University of Skövde P.O. 408, 541 48 Skövde,
More informationFast Efficient Clustering Algorithm for Balanced Data
Vol. 5, No. 6, 214 Fast Efficient Clustering Algorithm for Balanced Data Adel A. Sewisy Faculty of Computer and Information, Assiut University M. H. Marghny Faculty of Computer and Information, Assiut
More informationUsing a genetic algorithm for editing k-nearest neighbor classifiers
Using a genetic algorithm for editing k-nearest neighbor classifiers R. Gil-Pita 1 and X. Yao 23 1 Teoría de la Señal y Comunicaciones, Universidad de Alcalá, Madrid (SPAIN) 2 Computer Sciences Department,
More informationMultiobjective Job-Shop Scheduling With Genetic Algorithms Using a New Representation and Standard Uniform Crossover
Multiobjective Job-Shop Scheduling With Genetic Algorithms Using a New Representation and Standard Uniform Crossover J. Garen 1 1. Department of Economics, University of Osnabrück, Katharinenstraße 3,
More informationIncorporation of Scalarizing Fitness Functions into Evolutionary Multiobjective Optimization Algorithms
H. Ishibuchi, T. Doi, and Y. Nojima, Incorporation of scalarizing fitness functions into evolutionary multiobjective optimization algorithms, Lecture Notes in Computer Science 4193: Parallel Problem Solving
More informationSolving Multi-objective Optimisation Problems Using the Potential Pareto Regions Evolutionary Algorithm
Solving Multi-objective Optimisation Problems Using the Potential Pareto Regions Evolutionary Algorithm Nasreddine Hallam, Graham Kendall, and Peter Blanchfield School of Computer Science and IT, The Univeristy
More informationIn this paper, we investigate solving the data clustering problem through MOQPSO in order to find several possible
Quantum Behaved Particle Swarm Optimization for Data Clustering with Multiple Objectives Heyam Al-Baity Souham Meshoul Ata Kaban Computer Science Department Computer Science Department Computer Science
More informationMachine Learning (BSMC-GA 4439) Wenke Liu
Machine Learning (BSMC-GA 4439) Wenke Liu 01-25-2018 Outline Background Defining proximity Clustering methods Determining number of clusters Other approaches Cluster analysis as unsupervised Learning Unsupervised
More informationFinding Knees in Multi-objective Optimization
Finding Knees in Multi-objective Optimization Jürgen Branke 1, Kalyanmoy Deb 2, Henning Dierolf 1, and Matthias Osswald 1 1 Institute AIFB, University of Karlsruhe, Germany branke@aifb.uni-karlsruhe.de
More informationEvolving SQL Queries for Data Mining
Evolving SQL Queries for Data Mining Majid Salim and Xin Yao School of Computer Science, The University of Birmingham Edgbaston, Birmingham B15 2TT, UK {msc30mms,x.yao}@cs.bham.ac.uk Abstract. This paper
More informationGT HEURISTIC FOR SOLVING MULTI OBJECTIVE JOB SHOP SCHEDULING PROBLEMS
GT HEURISTIC FOR SOLVING MULTI OBJECTIVE JOB SHOP SCHEDULING PROBLEMS M. Chandrasekaran 1, D. Lakshmipathy 1 and P. Sriramya 2 1 Department of Mechanical Engineering, Vels University, Chennai, India 2
More informationMulti-objective Optimization
Jugal K. Kalita Single vs. Single vs. Single Objective Optimization: When an optimization problem involves only one objective function, the task of finding the optimal solution is called single-objective
More informationAdaptive Metric Nearest Neighbor Classification
Adaptive Metric Nearest Neighbor Classification Carlotta Domeniconi Jing Peng Dimitrios Gunopulos Computer Science Department Computer Science Department Computer Science Department University of California
More informationA New Crossover Technique for Cartesian Genetic Programming
A New Crossover Technique for Cartesian Genetic Programming Genetic Programming Track Janet Clegg Intelligent Systems Group, Department of Electronics University of York, Heslington York, YO DD, UK jc@ohm.york.ac.uk
More informationMultiobjective Prototype Optimization with Evolved Improvement Steps
Multiobjective Prototype Optimization with Evolved Improvement Steps Jiri Kubalik 1, Richard Mordinyi 2, and Stefan Biffl 3 1 Department of Cybernetics Czech Technical University in Prague Technicka 2,
More informationPerformance Assessment of DMOEA-DD with CEC 2009 MOEA Competition Test Instances
Performance Assessment of DMOEA-DD with CEC 2009 MOEA Competition Test Instances Minzhong Liu, Xiufen Zou, Yu Chen, Zhijian Wu Abstract In this paper, the DMOEA-DD, which is an improvement of DMOEA[1,
More informationFinding Sets of Non-Dominated Solutions with High Spread and Well-Balanced Distribution using Generalized Strength Pareto Evolutionary Algorithm
16th World Congress of the International Fuzzy Systems Association (IFSA) 9th Conference of the European Society for Fuzzy Logic and Technology (EUSFLAT) Finding Sets of Non-Dominated Solutions with High
More informationA Search Method with User s Preference Direction using Reference Lines
A Search Method with User s Preference Direction using Reference Lines Tomohiro Yoshikawa Graduate School of Engineering, Nagoya University, Nagoya, Japan, {yoshikawa}@cse.nagoya-u.ac.jp Abstract Recently,
More informationUnsupervised Feature Selection Using Multi-Objective Genetic Algorithms for Handwritten Word Recognition
Unsupervised Feature Selection Using Multi-Objective Genetic Algorithms for Handwritten Word Recognition M. Morita,2, R. Sabourin 3, F. Bortolozzi 3 and C. Y. Suen 2 École de Technologie Supérieure, Montreal,
More informationMulti-objective Optimization Algorithm based on Magnetotactic Bacterium
Vol.78 (MulGrab 24), pp.6-64 http://dx.doi.org/.4257/astl.24.78. Multi-obective Optimization Algorithm based on Magnetotactic Bacterium Zhidan Xu Institute of Basic Science, Harbin University of Commerce,
More informationAn Evolutionary Multi-Objective Crowding Algorithm (EMOCA): Benchmark Test Function Results
Syracuse University SURFACE Electrical Engineering and Computer Science College of Engineering and Computer Science -0-005 An Evolutionary Multi-Objective Crowding Algorithm (EMOCA): Benchmark Test Function
More informationAdaptive Multi-objective Particle Swarm Optimization Algorithm
Adaptive Multi-objective Particle Swarm Optimization Algorithm P. K. Tripathi, Sanghamitra Bandyopadhyay, Senior Member, IEEE and S. K. Pal, Fellow, IEEE Abstract In this article we describe a novel Particle
More informationApproximation-Guided Evolutionary Multi-Objective Optimization
Approximation-Guided Evolutionary Multi-Objective Optimization Karl Bringmann 1, Tobias Friedrich 1, Frank Neumann 2, Markus Wagner 2 1 Max-Planck-Institut für Informatik, Campus E1.4, 66123 Saarbrücken,
More informationThe Comparative Study of Machine Learning Algorithms in Text Data Classification*
The Comparative Study of Machine Learning Algorithms in Text Data Classification* Wang Xin School of Science, Beijing Information Science and Technology University Beijing, China Abstract Classification
More informationReference Point-Based Particle Swarm Optimization Using a Steady-State Approach
Reference Point-Based Particle Swarm Optimization Using a Steady-State Approach Richard Allmendinger,XiaodongLi 2,andJürgen Branke University of Karlsruhe, Institute AIFB, Karlsruhe, Germany 2 RMIT University,
More informationClustering. Robert M. Haralick. Computer Science, Graduate Center City University of New York
Clustering Robert M. Haralick Computer Science, Graduate Center City University of New York Outline K-means 1 K-means 2 3 4 5 Clustering K-means The purpose of clustering is to determine the similarity
More informationResearch on Applications of Data Mining in Electronic Commerce. Xiuping YANG 1, a
International Conference on Education Technology, Management and Humanities Science (ETMHS 2015) Research on Applications of Data Mining in Electronic Commerce Xiuping YANG 1, a 1 Computer Science Department,
More informationImproved S-CDAS using Crossover Controlling the Number of Crossed Genes for Many-objective Optimization
Improved S-CDAS using Crossover Controlling the Number of Crossed Genes for Many-objective Optimization Hiroyuki Sato Faculty of Informatics and Engineering, The University of Electro-Communications -5-
More informationDesign of Curves and Surfaces Using Multi-Objective Optimization
Design of Curves and Surfaces Using Multi-Objective Optimization Rony Goldenthal and Michel Bercovier Abstract. Design by optimization of curves and surfaces is a powerful design technique. The mathematical
More informationExploration of Pareto Frontier Using a Fuzzy Controlled Hybrid Line Search
Seventh International Conference on Hybrid Intelligent Systems Exploration of Pareto Frontier Using a Fuzzy Controlled Hybrid Line Search Crina Grosan and Ajith Abraham Faculty of Information Technology,
More informationA New Crossover Technique for Cartesian Genetic Programming
A New Crossover Technique for Cartesian Genetic Programming Genetic Programming Track Janet Clegg Intelligent Systems Group, Department of Electronics University of York, Heslington York,YODD,UK jc@ohm.york.ac.uk
More informationminimizing minimizing
The Pareto Envelope-based Selection Algorithm for Multiobjective Optimization David W. Corne, Joshua D. Knowles, Martin J. Oates School of Computer Science, Cybernetics and Electronic Engineering University
More informationMechanical Component Design for Multiple Objectives Using Elitist Non-Dominated Sorting GA
Mechanical Component Design for Multiple Objectives Using Elitist Non-Dominated Sorting GA Kalyanmoy Deb, Amrit Pratap, and Subrajyoti Moitra Kanpur Genetic Algorithms Laboratory (KanGAL) Indian Institute
More informationEvolutionary Multi-objective Optimization of Business Process Designs with Pre-processing
Evolutionary Multi-objective Optimization of Business Process Designs with Pre-processing Kostas Georgoulakos Department of Applied Informatics University of Macedonia Thessaloniki, Greece mai16027@uom.edu.gr
More informationExperimental Study on Bound Handling Techniques for Multi-Objective Particle Swarm Optimization
Experimental Study on Bound Handling Techniques for Multi-Objective Particle Swarm Optimization adfa, p. 1, 2011. Springer-Verlag Berlin Heidelberg 2011 Devang Agarwal and Deepak Sharma Department of Mechanical
More informationCluster Analysis for Microarray Data
Cluster Analysis for Microarray Data Seventh International Long Oligonucleotide Microarray Workshop Tucson, Arizona January 7-12, 2007 Dan Nettleton IOWA STATE UNIVERSITY 1 Clustering Group objects that
More informationImproved Crowding Distance for NSGA-II
Improved Crowding Distance for NSGA-II Xiangxiang Chu and Xinjie Yu Department of Electrical Engineering, Tsinghua University, Beijing84, China Abstract:Non-dominated sorting genetic algorithm II (NSGA-II)
More informationBinary Histogram in Image Classification for Retrieval Purposes
Binary Histogram in Image Classification for Retrieval Purposes Iivari Kunttu 1, Leena Lepistö 1, Juhani Rauhamaa 2, and Ari Visa 1 1 Tampere University of Technology Institute of Signal Processing P.
More informationdivision 1 division 2 division 3 Pareto Optimum Solution f 2 (x) Min Max (x) f 1
The New Model of Parallel Genetic Algorithm in Multi-Objective Optimization Problems Divided Range Multi-Objective Genetic Algorithm Tomoyuki HIROYASU Mitsunori MIKI Sinya WATANABE Doshisha University,
More informationET-based Test Data Generation for Multiple-path Testing
2016 3 rd International Conference on Engineering Technology and Application (ICETA 2016) ISBN: 978-1-60595-383-0 ET-based Test Data Generation for Multiple-path Testing Qingjie Wei* College of Computer
More informationOn The Effects of Archiving, Elitism, And Density Based Selection in Evolutionary Multi-Objective Optimization
On The Effects of Archiving, Elitism, And Density Based Selection in Evolutionary Multi-Objective Optimization Marco Laumanns, Eckart Zitzler, and Lothar Thiele ETH Zürich, Institut TIK, CH 8092 Zürich,
More informationDEMO: Differential Evolution for Multiobjective Optimization
DEMO: Differential Evolution for Multiobjective Optimization Tea Robič and Bogdan Filipič Department of Intelligent Systems, Jožef Stefan Institute, Jamova 39, SI-1000 Ljubljana, Slovenia tea.robic@ijs.si
More informationMultiobjective Optimization Using Adaptive Pareto Archived Evolution Strategy
Multiobjective Optimization Using Adaptive Pareto Archived Evolution Strategy Mihai Oltean Babeş-Bolyai University Department of Computer Science Kogalniceanu 1, Cluj-Napoca, 3400, Romania moltean@cs.ubbcluj.ro
More informationReducing Fitness Evaluations Using Clustering Techniques and Neural Network Ensembles
Reducing Evaluations Using Clustering Techniques and Neural Network Ensembles Yaochu Jin and Bernhard Sendhoff Honda Research Institute Europe Carl-Legien-Str. 30 63073 Offenbach/Main, Germany yaochu.jin@honda-ri.de
More informationTelecommunication and Informatics University of North Carolina, Technical University of Gdansk Charlotte, NC 28223, USA
A Decoder-based Evolutionary Algorithm for Constrained Parameter Optimization Problems S lawomir Kozie l 1 and Zbigniew Michalewicz 2 1 Department of Electronics, 2 Department of Computer Science, Telecommunication
More informationFuzzy-Pareto-Dominance and its Application in Evolutionary Multi-Objective Optimization
Fuzzy-Pareto-Dominance and its Application in Evolutionary Multi-Objective Optimization Mario Köppen, Raul Vicente-Garcia, and Bertram Nickolay Fraunhofer IPK, Pascalstr. 8-9, 10587 Berlin, Germany {mario.koeppen
More informationLeave-One-Out Support Vector Machines
Leave-One-Out Support Vector Machines Jason Weston Department of Computer Science Royal Holloway, University of London, Egham Hill, Egham, Surrey, TW20 OEX, UK. Abstract We present a new learning algorithm
More informationA Distance Metric for Evolutionary Many-Objective Optimization Algorithms Using User-Preferences
A Distance Metric for Evolutionary Many-Objective Optimization Algorithms Using User-Preferences Upali K. Wickramasinghe and Xiaodong Li School of Computer Science and Information Technology, RMIT University,
More informationA Web Page Recommendation system using GA based biclustering of web usage data
A Web Page Recommendation system using GA based biclustering of web usage data Raval Pratiksha M. 1, Mehul Barot 2 1 Computer Engineering, LDRP-ITR,Gandhinagar,cepratiksha.2011@gmail.com 2 Computer Engineering,
More informationEvolutionary Algorithms and the Cardinality Constrained Portfolio Optimization Problem
Evolutionary Algorithms and the Cardinality Constrained Portfolio Optimization Problem Felix Streichert, Holger Ulmer, and Andreas Zell Center for Bioinformatics Tübingen (ZBIT), University of Tübingen,
More informationComputing hypervolume contributions in low dimensions: asymptotically optimal algorithm and complexity results
EMO 2011, 5-8 April, Ouro Preto, Brazil Computing hypervolume contributions in low dimensions: asymptotically optimal algorithm and complexity results Michael T.M. Emmerich 1,2,3 and Carlos M. Fonseca
More informationMachine Learning (BSMC-GA 4439) Wenke Liu
Machine Learning (BSMC-GA 4439) Wenke Liu 01-31-017 Outline Background Defining proximity Clustering methods Determining number of clusters Comparing two solutions Cluster analysis as unsupervised Learning
More informationA Genetic Algorithm Applied to Graph Problems Involving Subsets of Vertices
A Genetic Algorithm Applied to Graph Problems Involving Subsets of Vertices Yaser Alkhalifah Roger L. Wainwright Department of Mathematical Department of Mathematical and Computer Sciences and Computer
More informationDevelopment of Evolutionary Multi-Objective Optimization
A. Mießen Page 1 of 13 Development of Evolutionary Multi-Objective Optimization Andreas Mießen RWTH Aachen University AVT - Aachener Verfahrenstechnik Process Systems Engineering Turmstrasse 46 D - 52056
More informationImage Classification for JPEG Compression
Image Classification for Compression Jevgenij Tichonov Vilnius University, Institute of Mathematics and Informatics Akademijos str. 4 LT-08663, Vilnius jevgenij.tichonov@gmail.com Olga Kurasova Vilnius
More informationEVOLUTIONARY algorithms (EAs) are a class of
An Investigation on Evolutionary Gradient Search for Multi-objective Optimization C. K. Goh, Y. S. Ong and K. C. Tan Abstract Evolutionary gradient search is a hybrid algorithm that exploits the complementary
More informationRobust Kernel Methods in Clustering and Dimensionality Reduction Problems
Robust Kernel Methods in Clustering and Dimensionality Reduction Problems Jian Guo, Debadyuti Roy, Jing Wang University of Michigan, Department of Statistics Introduction In this report we propose robust
More informationTexture Classification by Combining Local Binary Pattern Features and a Self-Organizing Map
Texture Classification by Combining Local Binary Pattern Features and a Self-Organizing Map Markus Turtinen, Topi Mäenpää, and Matti Pietikäinen Machine Vision Group, P.O.Box 4500, FIN-90014 University
More informationMulti-Objective Optimization using Evolutionary Algorithms
Multi-Objective Optimization using Evolutionary Algorithms Kalyanmoy Deb Department of Mechanical Engineering, Indian Institute of Technology, Kanpur, India JOHN WILEY & SONS, LTD Chichester New York Weinheim
More informationAn Application of Genetic Algorithm for Auto-body Panel Die-design Case Library Based on Grid
An Application of Genetic Algorithm for Auto-body Panel Die-design Case Library Based on Grid Demin Wang 2, Hong Zhu 1, and Xin Liu 2 1 College of Computer Science and Technology, Jilin University, Changchun
More informationImproving interpretability in approximative fuzzy models via multi-objective evolutionary algorithms.
Improving interpretability in approximative fuzzy models via multi-objective evolutionary algorithms. Gómez-Skarmeta, A.F. University of Murcia skarmeta@dif.um.es Jiménez, F. University of Murcia fernan@dif.um.es
More informationEFFECTIVE CONCURRENT ENGINEERING WITH THE USAGE OF GENETIC ALGORITHMS FOR SOFTWARE DEVELOPMENT
EFFECTIVE CONCURRENT ENGINEERING WITH THE USAGE OF GENETIC ALGORITHMS FOR SOFTWARE DEVELOPMENT D.Sundar 1, Dr.K.Alagarsamy 2 1 Assistant Professor, Thiagarajar School of Management, Madurai, India 2 Associate
More informationSemi-Supervised Clustering with Partial Background Information
Semi-Supervised Clustering with Partial Background Information Jing Gao Pang-Ning Tan Haibin Cheng Abstract Incorporating background knowledge into unsupervised clustering algorithms has been the subject
More informationRevision of a Floating-Point Genetic Algorithm GENOCOP V for Nonlinear Programming Problems
4 The Open Cybernetics and Systemics Journal, 008,, 4-9 Revision of a Floating-Point Genetic Algorithm GENOCOP V for Nonlinear Programming Problems K. Kato *, M. Sakawa and H. Katagiri Department of Artificial
More informationMulti-Objective Optimization using Evolutionary Algorithms
Multi-Objective Optimization using Evolutionary Algorithms Kalyanmoy Deb Department ofmechanical Engineering, Indian Institute of Technology, Kanpur, India JOHN WILEY & SONS, LTD Chichester New York Weinheim
More informationCommunity-Based Recommendations: a Solution to the Cold Start Problem
Community-Based Recommendations: a Solution to the Cold Start Problem Shaghayegh Sahebi Intelligent Systems Program University of Pittsburgh sahebi@cs.pitt.edu William W. Cohen Machine Learning Department
More informationClustering and Visualisation of Data
Clustering and Visualisation of Data Hiroshi Shimodaira January-March 28 Cluster analysis aims to partition a data set into meaningful or useful groups, based on distances between data points. In some
More informationNeural Network Weight Selection Using Genetic Algorithms
Neural Network Weight Selection Using Genetic Algorithms David Montana presented by: Carl Fink, Hongyi Chen, Jack Cheng, Xinglong Li, Bruce Lin, Chongjie Zhang April 12, 2005 1 Neural Networks Neural networks
More informationKBSVM: KMeans-based SVM for Business Intelligence
Association for Information Systems AIS Electronic Library (AISeL) AMCIS 2004 Proceedings Americas Conference on Information Systems (AMCIS) December 2004 KBSVM: KMeans-based SVM for Business Intelligence
More informationDynamic Ensemble Construction via Heuristic Optimization
Dynamic Ensemble Construction via Heuristic Optimization Şenay Yaşar Sağlam and W. Nick Street Department of Management Sciences The University of Iowa Abstract Classifier ensembles, in which multiple
More informationMechanical Component Design for Multiple Objectives Using Elitist Non-Dominated Sorting GA
Mechanical Component Design for Multiple Objectives Using Elitist Non-Dominated Sorting GA Kalyanmoy Deb, Amrit Pratap, and Subrajyoti Moitra Kanpur Genetic Algorithms Laboratory (KanGAL) Indian Institute
More informationAvailable online at ScienceDirect. Procedia Computer Science 60 (2015 )
Available online at www.sciencedirect.com ScienceDirect Procedia Computer Science 60 (2015 ) 178 187 19th International Conference on Knowledge Based and Intelligent Information and Engineering Systems
More informationBayesian Optimization Algorithms for Multi-Objective Optimization
Bayesian Optimization Algorithms for Multi-Objective Optimization Marco Laumanns 1 and Jiri Ocenasek 2 1 ETH Zürich, Computer Engineering and Networks Laboratory, CH 8092 Zürich laumanns@tik.ee.ethz.ch
More informationA Memetic Heuristic for the Co-clustering Problem
A Memetic Heuristic for the Co-clustering Problem Mohammad Khoshneshin 1, Mahtab Ghazizadeh 2, W. Nick Street 1, and Jeffrey W. Ohlmann 1 1 The University of Iowa, Iowa City IA 52242, USA {mohammad-khoshneshin,nick-street,jeffrey-ohlmann}@uiowa.edu
More informationImage Classification and Processing using Modified Parallel-ACTIT
Proceedings of the 2009 IEEE International Conference on Systems, Man, and Cybernetics San Antonio, TX, USA - October 2009 Image Classification and Processing using Modified Parallel-ACTIT Jun Ando and
More informationAssessing the Convergence Properties of NSGA-II for Direct Crashworthiness Optimization
10 th International LS-DYNA Users Conference Opitmization (1) Assessing the Convergence Properties of NSGA-II for Direct Crashworthiness Optimization Guangye Li 1, Tushar Goel 2, Nielen Stander 2 1 IBM
More informationA Fuzzy Logic Controller Based Dynamic Routing Algorithm with SPDE based Differential Evolution Approach
A Fuzzy Logic Controller Based Dynamic Routing Algorithm with SPDE based Differential Evolution Approach Debraj De Sonai Ray Amit Konar Amita Chatterjee Department of Electronics & Telecommunication Engineering,
More informationParallel Multi-objective Optimization using Master-Slave Model on Heterogeneous Resources
Parallel Multi-objective Optimization using Master-Slave Model on Heterogeneous Resources Author Mostaghim, Sanaz, Branke, Jurgen, Lewis, Andrew, Schmeck, Hartmut Published 008 Conference Title IEEE Congress
More informationREPRESENTATION OF BIG DATA BY DIMENSION REDUCTION
Fundamental Journal of Mathematics and Mathematical Sciences Vol. 4, Issue 1, 2015, Pages 23-34 This paper is available online at http://www.frdint.com/ Published online November 29, 2015 REPRESENTATION
More informationA Parallel Evolutionary Algorithm for Discovery of Decision Rules
A Parallel Evolutionary Algorithm for Discovery of Decision Rules Wojciech Kwedlo Faculty of Computer Science Technical University of Bia lystok Wiejska 45a, 15-351 Bia lystok, Poland wkwedlo@ii.pb.bialystok.pl
More informationMetaheuristic Development Methodology. Fall 2009 Instructor: Dr. Masoud Yaghini
Metaheuristic Development Methodology Fall 2009 Instructor: Dr. Masoud Yaghini Phases and Steps Phases and Steps Phase 1: Understanding Problem Step 1: State the Problem Step 2: Review of Existing Solution
More informationSanta Fe Trail Problem Solution Using Grammatical Evolution
2012 International Conference on Industrial and Intelligent Information (ICIII 2012) IPCSIT vol.31 (2012) (2012) IACSIT Press, Singapore Santa Fe Trail Problem Solution Using Grammatical Evolution Hideyuki
More informationLEARNING WEIGHTS OF FUZZY RULES BY USING GRAVITATIONAL SEARCH ALGORITHM
International Journal of Innovative Computing, Information and Control ICIC International c 2013 ISSN 1349-4198 Volume 9, Number 4, April 2013 pp. 1593 1601 LEARNING WEIGHTS OF FUZZY RULES BY USING GRAVITATIONAL
More informationMulti-Objective Memetic Algorithm using Pattern Search Filter Methods
Multi-Objective Memetic Algorithm using Pattern Search Filter Methods F. Mendes V. Sousa M.F.P. Costa A. Gaspar-Cunha IPC/I3N - Institute of Polymers and Composites, University of Minho Guimarães, Portugal
More informationDiscovering Knowledge Rules with Multi-Objective Evolutionary Computing
2010 Ninth International Conference on Machine Learning and Applications Discovering Knowledge Rules with Multi-Objective Evolutionary Computing Rafael Giusti, Gustavo E. A. P. A. Batista Instituto de
More information6. Concluding Remarks
[8] K. J. Supowit, The relative neighborhood graph with an application to minimum spanning trees, Tech. Rept., Department of Computer Science, University of Illinois, Urbana-Champaign, August 1980, also
More informationSPEA2: Improving the strength pareto evolutionary algorithm
Research Collection Working Paper SPEA2: Improving the strength pareto evolutionary algorithm Author(s): Zitzler, Eckart; Laumanns, Marco; Thiele, Lothar Publication Date: 2001 Permanent Link: https://doi.org/10.3929/ethz-a-004284029
More information